=Cell1-Cell2-Cell3-Cell4

returns zero (I believe in all versions of Excel), however

=(Cell1-Cell2-Cell3-Cell4)

returns 4.44089E-16. It is not an error, and it is not unique to Excel.

The latter result is the exact answer to an approximate problem. As Ron

Rosenfeld suggested, the issue is binary arithmetic. None of the values

in Cell2, Cell3, or Cell4 can be exactly represented in binary (which is

used by almost all general purpose computing software). Excel then does

exact arithmatic on the approximate inputs to get 4.44089E-16, which is

well within the advertised 15 figure accuracy.

The reason that the first form gets zero is not that the math is done

differently. Instead, when the final operation is a subtraction where

the result is zero to 15 figures, Excel arbitrarily zeroes the nonzero

result to try to avoid these kinds of questions. The second form

bypasses that fuzz factor because of the parenteses.

Your third equation

=Cell1-(Cell2+Cell3+Cell4)

still has to approximate the values in Cells 2 through 4, but the answer

to the approximate problem happens to be the same as the answer to your

decimal problem to the available precision.

Jerry